Aug 2026· Multimedia Systems· Vol 32· 0 citations· 51 references
TL;DR
This study proposes STaR, a novel retriever fine-tuning framework that integrates BM25 similarity graph-based soft labeling with a triplet similarity learning strategy based on Sentence-BERT (SBERT), and introduces a triplet-aware SBERT training architecture that explicitly models relative semantic distances between queries and candidate passages, significantly enhancing retrieval ranking precision and semantic robustness.
Evaluation on a multi-page technical PDF document shows that the hybrid retrieval and re-ranking stages together raise retrieval precision and reduce irrelevant or unsupported answers compared with retrieval limited to a single method, supporting the use of this approach for reliable, document-grounded question answering.
Vishwa K Dave, K. Pallavi· International Research Journ...· 0 citations
KBMR is proposed, the first MLLM-based embedding retriever tailored for KB-VQA, and an MLLM-based semantic discriminator that generates continuous entity-consistency weights is introduced to tackle the challenge of noisy supervision in Wikipedia-scale retrieval.
Hangrui Xu, Zhengxian Wu, Yu Yu et al.· 0 citations
MeAI++ is proposed, a novel framework that integrates knowledge graph based retrieval with a reinforcement learning (RL) optimization loop to jointly enhance retrieval and generation and confirms the effectiveness and generalizability of MeAI++ for complex, knowledge-intensive question answering.
Tram Nguyen, Truong H. V. Phan· Journal of Intelligent &...· 0 citations
Vision-language models (VLMs) augmented with retrieval-augmented generation (RAG) benefit from access to external evidence. However, standard retrievers and rerankers optimize for semantic similarity rather than answer utility, creating a preference gap: documents that appear relevant may not help the generator produce a correct answer. Motivated by this, we propose a two-stage generator-in-the-loop alignment framework that closes this gap without human document-level relevance annotations. Our framework consists of two stages: in Stage 1, a VLM generates a hypothetical text passage from the image-query pair, which is used as the retrieval query for dense text search, bridging the image-to-text modality gap. In Stage 2, a cross-encoder reranker adapted with low-rank adaptation (LoRA) is fine-tuned using answer-supervised preference pairs mined from the frozen VLM: given the dataset answer label, a candidate document is labeled positive if the VLM produces the correct answer when given that document as context, and negative otherwise. This generator-guided signal is compatible with multiple alignment loss functions, including contrastive (triplet) loss, pairwise direct preference optimization (DPO), and supervised fine-tuning (SFT), and supports periodic re-mining to refresh preference pairs as the reranker improves. Experiments on VQA-X and A-OKVQA with Qwen3.5-2B and Qwen3-VL-4B-Instruct show that our proposed framework consistently outperforms rank-order, random, and REPLUG-style likelihood baselines under various alignment losses and pool size settings, suggesting that answer-level generator feedback is an effective supervision signal for preference alignment.
Zhan-Lun Chang, Dong-Jun Han, Seyyedali Hosseinalipour et al.· 0 citations
Retrieval-Augmented Generation (RAG) is widely used to enhance question-answering systems across various domains. However, while real-world source documents are inherently structured, conventional RAG approaches primarily rely on semantic similarity between isolated text chunks, which can overlook document hierarchy and limit retrieval effectiveness. To address this issue, this study introduces a document hierarchy-based Chunk Graph approach to improve retrieval grounding in RAG systems. The proposed framework preserves document hierarchy during chunking and models inter-chunk relationships using a weighted graph that combines structural proximity and semantic similarity. The approach was evaluated using the StructuredQA and CUAD benchmark datasets, with performance measured via Precision, Recall, and F1-Score. Experimental results demonstrate that the effectiveness of the Chunk Graph depends heavily on the source document format. On the highly structured StructuredQA dataset, the proposed method successfully connects fragmented information, improving the F1-Score from 47.62% to 50.23%. Conversely, on the CUAD dataset which consists of raw text with implicit hierarchy and no nested structure, the model becomes redundant and does not yield performance gains. These findings conclude that integrating structural-semantic relationships significantly improves context selection, specifically for documents with explicit hierarchical structures.
P. Cristin, Hilmil Pradana· JOURNAL OF APPLIED INFORMATI...· 0 citations
DARE-RAG is proposed, an adaptive retrieval framework that activates LLM-based query expansion only for retrieval-challenging queries, and formulates expansion activation as a lightweight binary classification problem using probe retrieval signals, including score margin, variance, entropy, query length, and lexical specificity.
Li-Xiang Zhu· Poster Volume 0007 The 2026...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
